Harries et al. BMC Cancer 2010, 10:315 http://www.biomedcentral.com/1471-2407/10/315

RESEARCH ARTICLE Open Access AlterationsResearch article in LMTK2, MSMB and HNF1B expression are associated with the development of

Lorna W Harries*1, John RB Perry2, Paul McCullagh3 and Malcolm Crundwell4

Abstract Background: Genome wide association studies (GWAS) have identified several genetic variants that are associated with prostate cancer. Most of these variants, like other GWAS association signals, are located in non-coding regions of potential candidate , and thus could act at the level of the mRNA transcript. Methods: We measured the expression and isoform usage of seven prostate cancer candidate genes in benign and malignant prostate by real-time PCR, and correlated these factors with cancer status and genotype at the GWAS risk variants. Results: We determined that levels of LMTK2 transcripts in prostate adenocarcinomas were only 32% of those in benign tissues (p = 3.2 × 10-7), and that an independent effect of genotype at variant rs6465657 on LMTK2 expression in benign (n = 39) and malignant tissues (n = 21) was also evident (P = 0.002). We also identified that whilst HNF1B(C) and MSMB2 comprised the predominant isoforms in benign tissues (90% and 98% of total HNF1B or MSMB expression), HNF1B(B) and MSMB1 were predominant in malignant tissue (95% and 96% of total HNF1B or MSMB expression; P = 1.7 × 10-7 and 4 × 10-4 respectively), indicating major shifts in isoform usage. Conclusions: Our results indicate that the amount or nature of mRNA transcripts expressed from the LMTK2, HNF1B and MSMB candidate genes is altered in prostate cancer, and provides further evidence for a role for these genes in this disorder. The alterations in isoform usage we detect highlights the potential importance of alternative mRNA processing and moderation of mRNA stability as potentially important disease mechanisms.

Background early onset cases of prostate cancer (< 55yrs) are caused Cancer of the prostate is the most common male malig- by inherited factors [2], highlighting the importance of nancy in the Western world, accounting for 25% of all genetics in this disorder. Early-onset prostate cancer that male UK cancers (taken from http://info.cancerre- aggregates in families is likely to result from the inheri- searchuk.org/cancerstats). Although survival rates are tance of rare genetic variants that have a large impact. increasing, cancer of the prostate remains the second Several strong risk factors have been identified from fam- most common cause of cancer death in UK men after ily studies which include mutations in the ELAC2, RNA- lung cancer. Several risk factors have been identified, the SEL and MSR1 genes involved in immune response [4] main factors being age [1], family history [2] and ethnic and the BRCA1 and BRCA2 genes involved in the devel- origin [3]. Prostate cancer is uncommon in men under 50 opment of breast cancer [5]. years old, but 80% of men aged over 80 years were found Despite increases in the understanding of the familial to have cancerous cells in their prostate at the time of forms of prostate cancer, the aetiology of the remaining death [1]. Estimates suggest that between 30 - 40% of all 60% of cases remains unclear. The advent of genome wide association studies (GWAS) has resulted in a dramatic * Correspondence: [email protected] increase in the number of susceptibility loci that have 1 Institute of Biomedical and Clinical Sciences, Peninsula NIHR Clinical Research Facility, University of Exeter, Peninsula Medical School, Barrack Road, Exeter, been associated with the development of prostate cancer Devon, UK (table 1). Four independent studies in the Icelandic, UK Full list of author information is available at the end of the article

© 2010 Harries et al; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. Harries et al. BMC Cancer 2010, 10:315 Page 2 of 12 http://www.biomedcentral.com/1471-2407/10/315

Table 1: Loci that have been identified as significant determinants of prostate cancer risk in four independent genome wide association studies.

Gene Location Marker MAF Post replication P value Position within gene Source

HNF1B 17q21.3 rs4430796 0.479 9.58 × 10-10 IVS2 [7,8,34] rs11649743 0.479 1.7 × 10-9 IVS4

MSMB 10q11.2 rs10993994 0.339 8.7 × 10-29 5' UTR [8,9] 7.31 × 10-13

LMTK2 17q25.3 rs6465657 0.438 1.1 × 10-9 IVS9 [9]

SLC22A3 6q26-q27 rs9364554 0.233 5.5 × 10-10 IVS5 [9]

CTBP2 10q26.13 rs4962416 0.229 1.70 × 10-7 IVS1 [8]

MYEOV 11q13 rs10896449 0.492 1.76 × 10-9 Intergenic, 5' to gene [8]

JAZF1 7p15.2-p15.1 rs10486567 0.267 2.14 × 10-6 IVS2 [8] MAF = minor allele frequency in the HapMap CEU population. and American populations have identified more than 20 tissues, and whether their expression was influenced by distinct genomic locations as being implicated in suscep- genotype at the GWAS risk variants. Seven genes were tibility to prostate cancer [6-9]. selected for study (figure 1). We measured the total or Many of the GWAS association signals identified are isoform specific of the JAZF1 (JAZF zinc located in regions of the genome that are not translated finger 1), CTBP2 (C-terminal binding 2), LMTK2 into protein. This may indicate that the associated marker (lemur tyrosine kinase 2), SLC22A3 (solute carrier family is in close linkage disequilibrium with an uncharacterised 22 extraneuronal monoamine transporter, member 3), variant in the protein coding region, or that the func- MYEOV (myeloma overexpressed), MSMB (microsemi- tional SNP may cause its effect by epigenetic or histone noprotein beta) and HNF1B (HNF homeobox 1 beta) in a modification. The other possibility is that the variant may cohort of 39 non-malignant benign prostatic hyperplasia act at the level of the messenger RNA. (BPH) samples and 21 prostate adenocarcinoma samples, There are several mechanisms by which variants in and correlated gene expression with cancer status and non-coding regions of the mRNA transcripts could affect with the genetic variants identified in the GWAS (table gene function. Firstly, if the variant lies in the promoter or 1). We demonstrate two of the eight variants are corre- 5' untranslated region of the transcript, it may interfere lated with gene expression levels in non-malignant pros- with the efficiency of transcription or translation, as has tate tissues, and that the presence of adenocarcinoma is been shown to be the case for several human and rodent associated with altered gene expression levels and differ- genes [10-12]. Variants located within the intronic ential expression of alternatively processed isoforms for sequences may interfere with mRNA splicing [13]. The three of the candidate genes in prostate tissue itself. production of aberrant splice products has previously been shown to be very important in the aetiology of can- Methods cer in general [14,15], but may contribute to the biologi- Subject Details cal heterogeneity of prostate cancer in particular [16]. The prostate samples in this study were obtained from Finally, polymorphisms located in the 3' untranslated the Exeter tissue bank, and comprised prostate chippings region may disrupt regulatory elements necessary for taken during routine transurethral resection of the pros- mRNA stability, interfere with regulation of translation tate (TURP). Both cancer and BPH groups originated by microRNAs or influence the polyadenylation dynam- from a population of older men treated for urological ics of the transcript [17-19]. problems at the Royal Devon and Exeter Hospital, Exeter, In this series of experiments, we sought to investigate UK and were predominantly of Caucasian origin. Sam- whether the genes suggested by the GWAS for prostate ples were flash frozen in liquid nitrogen upon removal, cancer were differentially expressed in malignant prostate and stored at -80°C until required. Subjects with benign Harries et al. BMC Cancer 2010, 10:315 Page 3 of 12 http://www.biomedcentral.com/1471-2407/10/315

1 2 3 4 JAZF1 (NM_175061)

1 2 3 4 5 6 7 8 9 10 SLC22A3 (NM_021977)

1 2 3 4 5 6 7 8 9 10 11 12 13 LMTK2 (NM_014916)

1 2 MYEOV (NM_138768)

1 2 3 4 MSMB1 (NM_002443)

1 2 4 MSMB2 (NM_138634)

1 2 3 4 5 6 7 8 9 HNF1B(A) (NM_000458)

1 2 3 4 5 6 7 8 9 HNF1B(B) (NM_001165923)

1 2 3 4 HNF1B(C) (NM_006481)

1a 2 3 4 5 6 7 8 9 10 11 CTBP2(1) (NM_001083914)

2b 4 5 6 7 8 9 10 11 CTBP2(2) (NM_022802)

1b 2 3 4 5 6 7 8 9 10 11 CTBP2(3) (NM_001329)

1c 2 3 4 5 6 7 8 9 10 11 CTBP2(4) (AK290390 )

Figure 1 Schematic representation of the structure of transcripts studied here. The structure of the JAZF1, SLC22A3, LMTK2, MYEOV, MSMB, HNF1B and CTBP2 mRNAs are represented here, together with their RefSeq ID numbers. Exons are represented by numbered grey boxes. The black box represents the section of exon 3 absent from isoform HNF1B(B). The section of intronic sequence present in HNF1B(C) is indicated by a hatched box. prostatic hyperplasia (n = 39) had no sign of prostate and Gleason scoring by dissection of the TURP chips and malignancy at the time the samples were taken, and none subsequent histological examination of tissue from either had received a subsequent diagnosis of prostate cancer 5 side of the sample used for RNA analysis. All cancer sam- years following their diagnosis of benign prostatic hyper- ples were classified as prostatic adenocarcinomas, of plasia (BPH). The median age of the BPH cohort was 83 which 57% were Gleason score 6 or 7, and 43% were Glea- years (IQR 78,88) and the median age of the prostate can- son score 8-10. Further analysis of patient case notes cer cohort was 83 years (IQR 76,87). BPH samples were revealed that 8/19 prostate cancer patients and 6/37 used as the control group for several reasons; firstly, most Benign Prostatic Hyperplasia patients had been exposed males have evidence of BPH by the age of 70 or 80, so the to hormone treatment (Zoladex, Finasteride, Casodex or presence of some degree of BPH is 'normal' at the median Proscar). age at diagnosis in our prostate cancer cohort (77 yrs, [IQR 72,82]). Truly 'normal' prostate tissue would thus Ethics Statement only be obtained in a much younger control cohort, All participants had consented to their tissue being used which could introduce bias. Secondly, since collection of in this study, and ethical approval was obtained from the prostate tissue requires a surgical procedure, which South West Research Ethics Committee. This study was would only be undertaken in subjects with evidence of carried out according to the Declaration of Helsinki. symptoms of prostate enlargement. Even then, morpho- DNA extraction logically normal tissue may harbour premalignant genetic DNA was extracted from prostate chippings using the abnormalities that are not immediately obvious. How- Qiagen DNA mini kit (Qiagen, Crawley, UK) using stan- ever, we acknowledge that genetic heterogeneity between dard procedures on the fully automated QiaCube sample patients and controls may influence the gene expression preparation platform (Qiagen, Crawley, UK), following from individual to individual. Prostate cancer samples (n manual homogenisation of tissue. = 21) underwent confirmation of cancer status, histology Harries et al. BMC Cancer 2010, 10:315 Page 4 of 12 http://www.biomedcentral.com/1471-2407/10/315

P = 3.3 X 10-5 rs3801294 LMTK2 A:G ratio at

DNA cDNA control

Figure 2 Allele-specific expression of LMTK2 transcripts. This histogram demonstrates allele-specific differences in the expression of LMTK2 tran- scripts bearing either the G or A allele at rs3801294, which is in perfect linkage disequilibrium (r2 = 1; D' = 1) with the index SNP rs6465657. The sample identity is given on the X-axis and the ratio of transcripts carrying G alleles (marking the C allele of rs6465657) relative to those carrying the A allele (marking the T allele at rs6465657) is given on the Y-axis. Error bars represent the interquartile range (IQR) of the measurements. The level of signifi- cance as determined by pairwise Mann Whitney-U analyses is indicated by asterisks.

Choice of candidate genes Genotyping Seven candidate genes were identified for expression Genotyping was carried out by PCR amplification and analysis. These were chosen on several criteria. Firstly, we direct sequencing of the sequence flanking the variant in chose our candidate genes on the basis that the GWAS question; primers and conditions are available on request. index SNPs were located within the gene sequence itself, Amplification reactions contained 10 μl Megamix Royal or very close to the genes in question. The linkage dis- (Microzone, Haywards Heath, UK), 10 μM primers and equilibrium blocks identified in the GWAS may contain 50 ng DNA in a total volume of 20 μl. Sequence specific other genes (or variants) that may contribute to the risk of primers for each amplicon were tagged with 5' M13 tails developing prostate cancer, the majority of these are to allow sequencing to be performed with a universal located remotely to the association signal and do not rep- M13 primer. Single-strand sequencing was carried out resent better biological candidates. Secondly, all seven are using standard methods on an ABI 3730 (Applied Biosys- good biological candidates, having roles in cell prolifera- tems, Warrington, UK). Sequences were compared to a tion, regulation of apoptosis or prostate development and designated individual homozygous for the major allele of function. Analysis of the linkage disequilibrium structure the variant using Mutation Surveyor (version 3.20; Soft- around the GWAS index SNPs revealed several other Genetics, State College, PA). candidates, but these were not analysed due to a lack of expression in the prostate, a lack of convincing biological RNA extraction evidence for a role in prostate function or the presence of RNA was extracted from prostate chippings using the more plausible candidate genes. The remaining four sig- Qiagen RNAeasy kit (Qiagen, Crawley, UK) on the fully nals located on 3, 8, 19 and X were not automated QiaCube sample preparation platform (Qia- analysed due to the lack of an obvious candidate gene. gen, Crawley, UK), following manual homogenisation of tissue with an additional pass through a Qiagen Harries et al. BMC Cancer 2010, 10:315 Page 5 of 12 http://www.biomedcentral.com/1471-2407/10/315

Table 2: Total expression levels of candidate genes in non-malignant and prostate cancer tissues.

Gene Total Expression in non-malignant prostate tissues Total Expression in prostate cancer tissues Value

JAZF1 0.82 (1.7) 0.83 (0.97) 0.185

SLC22A3 1.01 (1.42) 0.48 (0.75) 0.015

LMTK2 1.10 (0.77) 0.35 (0.30) 2.4 × 10-7

MSMB 95.3 (326.3) 13.5 (90.9) 0.156

HNF1B 63.0 (106.8) 37.3 (56.0) 0.175

MYEOV 0.07 (0.15) 0.08 (0.15) 0.655

CTBP2 3.35 (2.54) 2.1 (1.6) 0.008 Gene expression levels are analysed relative to the beta 2 microglobulin (B2M) and beta glucorinidase (GUSB) genes, and normalised to the average value in the non-malignant samples. The interquartile range is given in parentheses. A P value of 0.007 (P = 0.05/n° of tests) was taken to indicate statistical significance following correction for multiple testing (7 independent tests). Statistically significant results are indicated in bold italic type.

Qiashreddder column (Qiagen, Crawley, UK) column to (ΔΔCt) method [20]. The abundance of each target in shear genomic DNA. RNA samples were stored at -80°C each individual was normalized to the average measure- prior to analysis. ment for the transcript in question across the cohort prior to correlation of expression levels with genotype. Reverse Transcription Approximately 500 ng of mRNA from each sample was LMTK2 Allele-specific expression treated with 1u of RNAse-free DNAse (Promega, Madi- Measurement of the relative abundance of transcripts son, USA) for 30 minutes at 37°C followed by 65°C for 10 bearing different alleles at variant rs3801294, in perfect minutes for nuclease inactivation. Complementary DNA linkage disequilibrium (r2 = 1; D' = 1) with the LMTK2 (cDNA) was synthesized from mRNA using the Super- index SNP rs6465657, was carried out in single tube reac- script III VILO RT-PCR system (Invitrogen, Paisley, UK) tions by allele-specific real time PCR. PCR amplicons according to manufacturer's instructions, with an incuba- were generated by a common set of primers for both tion temperature of 42°C and a reaction volume of 40 μl. alleles, but transcripts bearing different alleles were dis- criminated by the use of allele-specific probes labelled Quantitative real time PCR for the measurement of total or with different fluorochromes (6-fluorescein [6-FAM] and isoform-specific gene expression VIC); probes and primers available on request. This Real-time PCR reactions using target-specific probes allows a measurement of transcript abundance that is were carried out in triplicate using the ABI Prism 7900 independent of amplification efficiency. Real-time PCR HT platform (Applied Biosystems, Warrington, UK) and reactions using were carried out in triplicate using the contained 5 μl TaqMan Fast Universal Mastermix (no ABI Prism 7900 HT platform (Applied Biosystems, War- AMPerase) (Applied Biosystems, Warrington, UK), 0.9 rington, UK) and contained 5 μl TaqMan Fast PCR Mas- μM each primer, 0.25 μM probe and 2 μl cDNA reverse ter Mix, no AmpErase™ (Applied Biosystems, transcribed as above in a total volume of 10 μl. PCR con- Warrington, UK), 72 μM each primer and 16 μM each ditions were a single cycle of 95°C for 20 seconds followed probe) in a total volume of 10 μl. PCR conditions were a by 50 cycles of 95°C for 1 second and 60°C for 20 seconds. single cycle of 95°C for 20 seconds followed by 60 cycles Individual probe and primer sets were pre-validated off- of 95°C for 1 second and 60°C for 20 seconds. The expres- the-shelf assays to specific transcripts provided by sion level of transcripts bearing the G allele (marking the Applied Biosystems (Warrington, UK; Assay identifica- C allele at rs6465657) was then calculated relative to that tion numbers available on request). The relative expres- of transcripts bearing the A allele (marking the T allele at sion level of each isoform was then determined relative to rs6465657) by the comparative (ΔΔCt) method [20], as the B2 M and GUSB transcripts by the comparative we have done for several previous studies [21,22]. Quanti- Harries et al. BMC Cancer 2010, 10:315 Page 6 of 12 http://www.biomedcentral.com/1471-2407/10/315

P = 2.4 X 10-7

BPH Prostate cancer

Figure 3 Expression of LMTK2 transcripts is reduced in prostate cancer. This boxplot gives the cancer status (benign prostatic hyperplasia or prostate adenocarcinoma) on the X-axis, and the expression of the single isoform of the LMTK2 gene relative to the endogenous controls beta-2-mi- croglobulin (B2M) and beta-glucorinidase (GUSB) on the Y-axis. Error bars represent the interquartile range (IQR) of the measurements. The level of significance as determined by linear regression of natural logarithmically transformed data is indicated. Outliers are indicated by circles. fications were then compared to the average measure- mens and 21 prostate adenocarcinoma samples (310 for ment obtained from eleven independent DNA samples, the BPH cohort and 116 for the cancer cohort). All three which should provide a known 50:50 mix of A and G- genotypes were detected for the variants, with the excep- bearing transcripts, in order to control for differences in tion of minor allele (C) homozygotes for rs4962416 probing efficiencies. within the CTBP2 gene, as expected given the minor allele frequency of only 0.229 in the HapMap CEU popu- Statistics lation. The statistical significance of apparent differences in gene expression levels was investigated by Kruskal Wallis anal- Risk genotypes at rs6465657 and rs4962416 may be ysis for multiway comparisons and by Mann Whitney-U associated with differences in LMTK2 and CTBP2 analysis for pairwise comparisons. Non-parametric sta- expression levels in benign prostatic hyperplasia samples tistics were employed due to the relatively small sample Analysis of potential correlations between genotype at numbers and the fact that the data obtained were not the GWAS association loci and isoform-specific gene normally distributed. In the case of the association expression of nearby candidate genes revealed correla- between LMTK2 expression and cancer status, linear tions between risk genotype at the GWAS variants regression was carried out on natural logarithmically rs6465657 and rs4962416 with the expression of LMTK2 transformed data to allow adjustment for the effect of and CTBP2 respectively. Individuals carrying two copies genotype at rs6465657, which we found to be indepen- of the protective T allele of rs6465657 expressed almost a dently associated with LMTK2 expression. All statistical third less LMTK2 mRNA than did individuals carrying analyses were carried out using SPSS v15.0 (SSPS PLC, one or two copies of the risk 'C' allele (0.68 [Interquartile Chicago, USA). range [IQR] = 0.63] v/s 1.18 [IQR = 0.56]; P = 0.007). Sim- ilarly, individuals homozygous for the 'T' allele at Results rs4962416 expressed 37% less of a partially characterised Genotyping of GWAS risk loci in benign prostatic isoform of CTBP2; CTBP2(4), than did individuals with hyperplasia and prostate cancer samples CT genotypes (0.82 [IQR 0.83] v/s 1.31 [IQR 0.96]; P = We obtained 426/480 genotypes for the 8 GWAS variants 0.009). Although neither of these findings quite reaches in our cohort of 39 benign prostatic hyperplasia speci- full statistical significance following adjustment for mul- Harries et al. BMC Cancer 2010, 10:315 Page 7 of 12 http://www.biomedcentral.com/1471-2407/10/315

Table 3: Isoform-specific expression levels of candidate genes in non-malignant and prostate cancer tissues.

Gene Total Expression in non-malignant prostate tissues Total Expression in prostate cancer tissues value

MSMB1 1.8 (6.9) 16.2 (85.2) 4 × 10-4

MSMB2 125.3 (409.8) 0.6 (4.73) 5.8 × 10-5

HNF1B(A) 1.2 (2.7) 0.94(1.52) 0.66

HNF1B(B) 5.1 (6.8) 35.1 (52.3) 1.7 × 10-7

HNF1B(C) 51.8 (109.9) 12 (3.4) 3.9 × 10-8

CTBP2(1) 1.3 (1.11) 0.62 (0.6) 0.009

CTBP2(3) 0.62 (0.55) 0.45 (0.58) 0.127

CTBP2(4) 1.4 (0.86) 0.70 (0.73) 0.007 Gene expression levels are analysed relative to the beta 2 microglobulin (B2M) and beta glucorinidase (GUSB) genes, and normalised to the average value in the non-malignant samples. The interquartile range is given in parentheses. A P value of 0.006 (P = 0.05/n° of tests) was taken to indicate statistical significance following correction for multiple testing (8 independent tests). Statistically significant results are indicated in bold italic type. tiple testing (P = < 0.004 for 12 independent tests), they with the presence of prostate cancer. Tumour samples do indicate that these genes are worthy of further study. expressed 68% less LMTK2 mRNA than did non-malig- nant BPH samples (figure 3; relative expression levels Allele-specific differences in the expression of LMTK2 were 1.10 [IQR 0.77] in BPH versus 0.35 [IQR 0.30] in transcripts bearing different alleles of rs6465657 prostate cancer; p = 2.4 × 10-7), after correction for the We measured the abundance of each LMTK2 allele of potential effect of genotype at rs6465657. Amongst the rs6465657 in 17 heterozygous individuals, and deter- eight members of the prostate cancer cohort who had mined that transcripts bearing the rs3801294 'A' allele, exposure to antiandrogenic drugs, no effect of prior hor- which is in perfect linkage disequilibrium (r2 = 1; D' = 1) mone treatment on the expression levels of LMTK2 was with the protective 'T' allele at rs6465657, were signifi- noted (0.49 [IQR 0.29) for untreated subjects versus 0.32 cantly less abundant in heterozygous BPH tissues than [IQR 0.30] in treated subjects; P = 0.270). were those carrying the 'G' allele, linked to the 'C' allele of We also noted differences in the total expression of the rs6465657. The ratio of protective 'A' alleles to risk 'G' SLC22A3 and CTBP2 genes in BPH compared with pros- alleles in heterozygous individuals is 0.65 [IQR 0.12]:1 tate cancer. CTBP2 and SLC22A3 mRNA levels were [IQR 0.33]; P = 3.3 × 10-5, figure 2), which remains statis- reduced by 27% and 50% in prostate cancer tissues com- tically significant after adjustment for multiple testing. pared with BPH tissues (4.2 [IQR 2.5] v/s 2.1[IQR 1.6]; P This provides further evidence for a role of LMTK2 in = 0.008 for CTBP2 and 0.48 [IQR 0.75] v/s 1.01 [IQR determining susceptibility to prostate cancer. No correla- 1.42]; P = 0.015 for SLC22A3). CTBP2(1) and CTBP2(4) tions of genotype with expression levels were noted for transcripts showed a trend towards lower expression in the remaining isoforms of the CTBP2 gene, or any iso- prostate cancer subjects treated with hormone therapy form of the JAZF1, MYEOV, SLC22A3, HNF1B or MSMB (CTBP2(1) ; (0.87 [0.40] in untreated subjects versus genes in BPH tissues. 0.31[IQR0.17]; p = 0.006) and CTBP2(4) transcripts (1.01[IQR 0.99] in untreated subjects versus 0.45 Total LMTK2 expression is altered in prostate cancer tissues [IQR0.42] in treated subjects; p = 0.006) but these results compared with non-malignant BPH samples do not reach statistical significance following adjustment We correlated expression levels at each of the loci with for multiple testing (P = < 0.004 for 12 independent tests). prostate cancer status (table 2). We found that expression This observation was also not noted in the BPH cohort (P levels of the LMTK2 gene were also inversely correlated Harries et al. BMC Cancer 2010, 10:315 Page 8 of 12 http://www.biomedcentral.com/1471-2407/10/315

4.0 X 10-10

BPH Prostate cancer

Figure 4 The ratio of alternatively-expressed isoforms of the MSMB gene is altered in prostate cancer tissues. This boxplot gives the cancer status (benign prostatic hyperplasia or prostate adenocarcinoma) on the X-axis, and the ratio of MSMB1:MSMB2 isoform expression calculated relative to endogenous controls beta-2-microglobulin (B2M) and beta-glucorinidase (GUSB) on the Y-axis. Error bars represent the interquartile range (IQR) of the measurements. The level of significance as determined by pairwise Mann Whitney-U analyses is indicated. Outliers are indicated by circles.

= 0.77 and 0.69 for CTBP2(1) and CTBP2(4) respectively. this tissue type. However, in cancer tissues, there is a near No association of cancer status with expression level was complete switch to isoform MSMB1, which is present at found for a control gene (SFRS protein kinase 1; SRPK1), 96% total MSMB expression in prostate cancer. The ratio not known to be associated with prostate cancer in any of MSMB1:MSMB2 isoforms is altered from 0.02 [IQR GWAS. We did not identify any correlations of total gene 0.02]: 1 in BPH tissue to 25.7 [IQR 30.2]: 1 in prostate expression with cancer status for the JAZF1, MSMB, adenocarcinoma tissues (P = 4.0 × 10-10, figure 4). In the HNF1B, MYEOV or CTBP2 genes. No correlations with case of the HNF1B gene, coding for 3 isoforms, we noted transcript expression were noted for cancer stage a similar switch in isoform expression. The predominant (assessed by Gleason grading of the cancer) or with expo- isoforms in BPH samples is HNF1B(C), present at 90% of sure to antiandrogenic drugs for any other transcripts total HNF1B expression. In prostate cancer tissues how- analysed in either prostate cancer or BPH cohorts. ever, isoform HNF1B(B) is predominant at 95% total HNF1B expression, with the levels of HNF1B(C) dropping Alterations to the relative balance of alternatively spliced to only 3%. The proportion of isoform HNF1B(A) does forms of the MSMB and HNF1B genes in prostate cancer not change. The ratio of HNF1B(B):HNF1B(C) isoforms tissues changes from 0.7 [IQR 0.22]: 1 in BPH to 35.4 [IQR Although we found no differences in total HNF1B, 124.71]: 1 in prostate cancer tissues (P = 2.9 × 10-9; figure CTBP2 or MSMB expression between BPH and prostate 5). We also found 47% and 50% reductions in CTBP2(1) cancer cases, these genes are alternatively processed. We and CTBP2(4) expression levels in prostate cancer tissue therefore sought to determine if there were differences in compared with benign prostate tissues (P = 0.009 and the relative balance of isoforms within our sample cohort 0.007 for CTBP2(1) and CTBP2(4) respectively), but these (table 3). We found highly significant disturbances to the values do not reach statistical significance following profile of isoforms expressed in prostate cancer in the adjustment for multiple testing (P = < 0.003 for 12 tests). case of the MSMB and HNF1B genes. In BPH tissues, the Previous exposure to antiandrogenic drugs did not affect primary isoform expressed at the MSMB locus is the ratio of MSMB or HNF1B splice variants in samples MSMB2, which comprises 98% of MSMB expression in from prostate cancer or BPH patients (P = 0.24 in Harries et al. BMC Cancer 2010, 10:315 Page 9 of 12 http://www.biomedcentral.com/1471-2407/10/315

2.9 X 10-9

BPH Prostate cancer

Figure 5 The ratio of alternatively-expressed isoforms of the HNF1B gene is altered in prostate cancer tissues. This boxplot gives the cancer status (benign prostatic hyperplasia or prostate adenocarcinoma) on the X-axis, and the ratio of HNF1B(B):HNF1B(C) isoform expression calculated rel- ative to endogenous controls beta-2-microglobulin (B2M) and beta-glucorinidase (GUSB) on the Y-axis. Error bars represent the interquartile range (IQR) of the measurements. The level of significance as determined by pairwise Mann Whitney-U analyses is indicated. Outliers are indicated by circles. untreated subjects and 0.69 in treated subjects for pros- The precise identity and mode of action of the functional tate cancer cohort and P = 0.88 24 in untreated subjects variant tagged by rs6465657 is at present unknown, but It and 0.72 in treated subjects for the BPH cohort). is reasonably easy to postulate that the 'C' allele at rs6465657 could be marking a loss-of-function variant Discussion with effects on LMTK2 half-life or function. We report here that disruption of the amount or nature of The LMTK2 gene codes for a transmembrane serine/ transcripts expressed from the LMTK2, MSMB and threonine/tyrosine kinase, with a role in endosomal HNF1B genes, identified in the genome wide scans for membrane trafficking [23]. It is also associated with prostate cancer, may be important in the aetiology of this NGF-TrkA signalling in murine brain, where it is a nega- disorder. tive regulator of NGF-induced neuronal differentiation We identified a 68% reduction in the expression of the [24]. LMTK2 interacts negatively with several other pro- LMTK2 gene in prostate tissue with evidence of the pres- teins with roles in cell division, such as protein phos- ence of adenocarcinoma when compared with non- phatase-1 (PP1C) and Inhibitor-2 (Inh2), which are part malignant BPH samples. Expression levels were 1.10 [IQR of a complex regulating separation of centrosomes during 0.77] in BPH versus 0.35 [IQR 0.30] in prostate cancer; p mitosis [25], and the cyclin-dependent kinase 5 (cdk5)/ = 3.2 × 10-7). We also noted an effect of genotype at the p35 complex [26], which has several functions, including LMTK2 variant rs6465657, identified in the GWAS as a a role in cell cycle progression. Recent studies have sug- susceptibility factor for prostate cancer whereby individ- gested that LMTK2 may interact with myosin IV, which uals carrying two alleles of the protective 'T' allele allelic has been shown to regulate both prostate specific antigen status expressed almost a third less LMTK2 than did indi- (PSA) and vascular endothelial growth factor (VEGF)[27], viduals carrying one or more 'C' alleles (P = 0.002). We both of which are associated with cancer. It is therefore suggest that this finding probably arises from a potential likely that that reduction in the amount or activity of imbalance in the transcription or stability of rs6465657 LMTK2 may lead to an increase in the proliferative alleles that we note upon allele-specific PCR (figure 1). capacity of prostate cells. Harries et al. BMC Cancer 2010, 10:315 Page 10 of 12 http://www.biomedcentral.com/1471-2407/10/315

We also report a potential role for alternative mRNA predict that the alterations to HNF1B profile we note may processing of the HNF1B and MSMB genes in the aetiol- manifest as an alteration to the overall activity of the ogy of prostate cancer. MSMB codes for a secreted semi- HNF1B gene and/or activation of a variant set of its target noprotein, which has tumour suppressor properties and genes. is thought to be silenced in prostate tumour tissues by the Our study provides good evidence that some of the enhancer of zeste homolog 2 (EZH2) protein [28]. GWAS associations for prostate cancer may be attributed Accordingly, MSMB expression has previously been to mRNA effects, but does have a number of caveats. We reported to be a positive prognostic indicator in prostate have chosen a single gene for analysis in the case of each cancer [29], although both these studies measured total, GWAS variant. The variants may tag a large region con- not isoform-specific expression levels. The MSMB gene taining many genes, some of which may also be good can- produces 2 isoforms; MSMB1 and MSMB2, which arise didates. Other genes in the same linkage disequilibrium from the skipping of exon 3 in MSMB2. This causes a (D') blocks, or even further distant, could also be impor- frameshift effect leading to the production of 2 distinct tant. This was recently documented for the eye colour . Previous studies suggest that both MSMB1 and trait in humans, where variant rs12913832 located within MSMB2 are present in normal prostate and normal gas- intron 86 of the HERC2 gene exerts its effects not by tric mucosa, but that MSMB2 is absent from the majority moderating HERC2 activity, but by moderating the effi- of a small series of gastric and prostatic carcinomas [30]. ciency of the promoter of the neighboring OCA2 gene, This is mostly in agreement with our findings, which sug- located 21Kbp upstream [37]. We have attempted to gest a major shift in MSMB expression in association with ensure we have chosen the correct gene on the basis of prostate cancer. We found MSMB2 to be the predomi- proximity and biological function, but other genes may nant isoform in benign prostate tissue, although small also be affected. Our analysis was also based on compari- amounts (~2%) of MSMB1 were also present. In prostate son of samples from patients with prostate cancer, and adenocarcinoma tissues, however, MSMB expression samples from patients with non-malignant Benign Pros- derived almost completely from MSMB1, which is pres- tatic Hyperplasia (BPH), rather than a paired analysis due ent at 96% of total MSMB expression. These findings may to the difficulty in obtaining normal prostate tissue from indicate that the tumour suppressor properties of MSMB the cancer patients. Our results may thus be influenced are derived from isoform MSMB2, and that MSMB1 iso- by factors both within and between groups, such as eth- forms are potentially pro-carcinogenic. This situation is nic origin or the use of antiandrogenic drugs. No differ- not uncommon for alternatively spliced genes in cancer; ences in the ethnic makeup of the cancer and control the vascular endothelial growth factor (VEGF) gene has cohorts were noted, with both being almost exclusively of previously been shown to code for both pro- and anti- Caucasian origin. Similarly, treatment with antiandro- angiogenic isoforms, with different behaviours in tumour gens had a minimal impact on the expression of any of the tissues [16]. transcripts studied; with the exception of CTBP2(1) and The HNF1B gene, which encodes three isoforms, CTBP2(4) transcripts, which showed a trend towards HNF1B(A), HNF1B(B) and HNF1B(C), in humans [31], lower expression in samples from prostate cancer demonstrates a similar alteration to the relative balance patients treated with hormone therapy, although this of alternatively expressed isoforms associated with cancer result was not statistically significant and no correlations of the prostate. The predominant isoform switches from of the expression of either transcript with hormone ther- HNF1B(C), which is expressed in non-malignant prostate apy was noted in the BPH cohort. No effect of hormone tissues, to HNF1B(B) , which is expressed in prostate ade- therapy on LMTK2 expression, or on the ratio of MSMB nocarcinoma samples. HNF1B is a transcription factor or HNF1B transcripts was noted. expressed in a limited number of tissues, and has previ- ously been associated with renal and ovarian tumours Conclusions [32,33] as well as prostate cancer [34]. The HNF1B iso- In this study, we suggest that the overall mRNA expres- forms are known to exhibit differences in function and sion level and/or the relative balance of alternatively target specificities; Isoforms (A) and (B) are transcrip- expressed isoforms of the LMTK2, MSMB and HNF1B tional activators whereas isoform HNF1B(C) is a tran- genes may be important determinants in the develop- scriptional repressor [35]. HNF1B isoforms have also ments of prostate cancer, and demonstrate the impor- been reported to activate different targets, for example, tance of alternative mRNA processing mechanisms such the HNF1B(C) isoform specifically has been demon- as alternative splicing or differential use of polyadenyla- strated to negatively regulate the Glutahione-S-trans- tion sites in gene regulation. Our studies highlight a clear ferase A (GSTA) promoter, via a mechanism that involves need for RNA studies to complement the genome wide Il-1beta [36]. Given the difference in the transcriptional association studies for prostate cancer and other diseases. properties and target specificity of these isoforms, we Harries et al. BMC Cancer 2010, 10:315 Page 11 of 12 http://www.biomedcentral.com/1471-2407/10/315

Abbreviations newly identified loci associated with prostate cancer susceptibility. GWAS: Genome wide association study, SNP: Single nucleotide polymorphism, Nat Genet 2008, 40(3):316-321. TURP: Transurethral resection of the prostate, BPH: Benign prostatic hyperpla- 10. Ord T, Ord D, Koivomagi M, Juhkam K, Ord T: Human TRB3 is upregulated sia, mRNA: Messenger RNA, RTPCR: Reverse transcription polymerase chain in stressed cells by the induction of translationally efficient mRNA reaction, 5' UTR: 5' Untranslated region, 3' UTR: 3' Untranslated region. containing a truncated 5'-UTR. Gene 2009, 444(1-2):24-32. 11. Ishii H, Kobayashi M, Sakuma Y: Alternative promoter usage and Competing interests alternative splicing of the rat estrogen receptor alpha gene generate The authors declare that they have no competing interests. numerous mRNA variants with distinct 5'-ends. J Steroid Biochem Mol Biol 2010, 118(1-2):59-69. Authors' contributions 12. Welham SJ, Clark AJ, Salter AM: A novel liver specific isoform of the rat LWH designed and executed the study, analysed results and drafted the manu- LAR transcript is expressed as a truncated isoform encoded from a script. JRBP advised on polygenic aspects of the study and reviewed the manu- 5'UTR located within intron 11. BMC Mol Biol 2009, 10:30. script. MC provided BPH and prostate adenocarcinoma samples and PM 13. Cartegni L, Chew SL, Krainer AR: Listening to silence and understanding carried out the histological examination of tissues to prove malignancy. All nonsense: exonic mutations that affect splicing. Nat Rev Genet 2002, authors read and approved the final submission. 3(4):285-298. 14. Fackenthal JD, Godley LA: Aberrant RNA splicing and its functional Acknowledgements consequences in cancer cells. Dis Model Mech 2008, 1(1):37-42. We are grateful to the Northcott Devon Medical Foundation [grant number TB/ 15. Pajares MJ, Ezponda T, Catena R, Calvo A, Pio R, Montuenga LM: MG/NO5002] for funding the study. We would like to thank the Peninsula NIHR Alternative splicing: an emerging topic in molecular and clinical Clinical Research Facility and the Exeter Tissue Bank for provision of BPH and oncology. Lancet Oncol 2007, 8(4):349-357. prostate cancer samples. We would like to acknowledge Albert Hallsworth and 16. Rajan P, Elliott DJ, Robson CN, Leung HY: Alternative splicing and Dr. Bridget Knight from Peninsula NIHR Clinical Research Facility for their excel- biological heterogeneity in prostate cancer. Nat Rev Urol 2009, lent technical assistance, and particularly Neelima Bhupathiraju and Rob Bolt 6(8):454-460. for help in gathering additional clinical details on the cohorts studied. We also 17. Le Quesne JP, Spriggs KA, Bushell M, Willis AE: Dysregulation of protein thank Professor Tim Frayling for critical appraisal of the manuscript. synthesis and disease. J Pathol 2010, 220(2):140-51. 18. Chatterjee S, Pal JK: Role of 5'- and 3'-untranslated regions of mRNAs in human diseases. Biol Cell 2009, 101(5):251-262. Author Details 19. Chen JM, Ferec C, Cooper DN: A systematic analysis of disease- 1Institute of Biomedical and Clinical Sciences, Peninsula NIHR Clinical Research associated variants in the 3' regulatory regions of human protein- Facility, University of Exeter, Peninsula Medical School, Barrack Road, Exeter, coding genes I: general principles and overview. Hum Genet 2006, Devon, UK, 2Genetics of Complex Traits, Peninsula Medical School, University of 120(1):1-21. Exeter, Exeter, UK, 3Department of Histopathology, Royal Devon and Exeter 20. Applied-Biosystems: Relative quantitation of gene expression. User NHS Foundation Trust, Exeter, UK and 4Department of Urology, Royal Devon bulletin #2 2001:11-15. and Exeter NHS Foundation Trust, Exeter, UK 21. Harries LW, Bingham C, Bellanne-Chantelot C, Hattersley AT, Ellard S: The Received: 4 March 2010 Accepted: 22 June 2010 position of premature termination codons in the hepatocyte nuclear Published: 22 June 2010 factor -1 beta gene determines susceptibility to nonsense-mediated This©BMC 2010 isarticleCancer an Harries Open is 2010, available et Access al;10 licensee:315 from:article http: BioMeddistributed//www.biomedcentral.com/1471-2407/10/315 Central under Ltd. the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. decay. Hum Genet 2005, 118(2):214-224. 22. Harries LW, Wickham CL, Evans JC, Rule SA, Joyner MV, Ellard S: Analysis of References haematopoietic chimaerism by quantitative real-time polymerase 1. Sakr WA, Grignon DJ, Haas GP, Heilbrun LK, Pontes JE, Crissman JD: Age chain reaction. Bone Marrow Transplant 2005, 35(3):283-290. and racial distribution of prostatic intraepithelial neoplasia. Eur Urol 23. Inoue T, Kon T, Ohkura R, Yamakawa H, Ohara O, Yokota J, Sutoh K: BREK/ 1996, 30(2):138-144. LMTK2 is a myosin VI-binding protein involved in endosomal 2. Bratt O: Hereditary prostate cancer: clinical aspects. J Urol 2002, membrane trafficking. Genes Cells 2008, 13(5):483-495. 168(3):906-913. 24. Kawa S, Fujimoto J, Tezuka T, Nakazawa T, Yamamoto T: Involvement of 3. Ben-Shlomo Y, Evans S, Ibrahim F, Patel B, Anson K, Chinegwundoh F, BREK, a serine/threonine kinase enriched in brain, in NGF signalling. Corbishley C, Dorling D, Thomas B, Gillatt D, et al.: The risk of prostate Genes Cells 2004, 9(3):219-232. cancer amongst black men in the United Kingdom: the PROCESS 25. Eto M, Elliott E, Prickett TD, Brautigan DL: Inhibitor-2 regulates protein cohort study. Eur Urol 2008, 53(1):99-105. phosphatase-1 complexed with NimA-related kinase to induce 4. De Marzo AM, Platz EA, Sutcliffe S, Xu J, Gronberg H, Drake CG, Nakai Y, centrosome separation. J Biol Chem 2002, 277(46):44013-44020. Isaacs WB, Nelson WG: Inflammation in prostate carcinogenesis. Nat Rev 26. Kesavapany S, Lau KF, Ackerley S, Banner SJ, Shemilt SJ, Cooper JD, Leigh Cancer 2007, 7(4):256-269. PN, Shaw CE, McLoughlin DM, Miller CC: Identification of a novel, 5. Mitra A, Fisher C, Foster CS, Jameson C, Barbachanno Y, Bartlett J, Bancroft membrane-associated neuronal kinase, cyclin-dependent kinase 5/ E, Doherty R, Kote-Jarai Z, Peock S, et al.: Prostate cancer in male BRCA1 p35-regulated kinase. J Neurosci 2003, 23(12):4975-4983. and BRCA2 mutation carriers has a more aggressive phenotype. Br J 27. Puri C, Chibalina MV, Arden SD, Kruppa AJ, Kendrick-Jones J, Buss F: Cancer 2008, 98(2):502-507. Overexpression of myosin VI in prostate cancer cells enhances PSA and 6. Duggan D, Zheng SL, Knowlton M, Benitez D, Dimitrov L, Wiklund F, VEGF secretion, but has no effect on endocytosis. Oncogene Robbins C, Isaacs SD, Cheng Y, Li G, et al.: Two genome-wide association 29(2):188-200. studies of aggressive prostate cancer implicate putative prostate 28. Beke L, Nuytten M, Van Eynde A, Beullens M, Bollen M: The gene tumor suppressor gene DAB2IP. J Natl Cancer Inst 2007, encoding the prostatic tumor suppressor PSP94 is a target for 99(24):1836-1844. repression by the Polycomb group protein EZH2. Oncogene 2007, 7. Gudmundsson J, Sulem P, Steinthorsdottir V, Bergthorsson JT, 26(31):4590-4595. Thorleifsson G, Manolescu A, Rafnar T, Gudbjartsson D, Agnarsson BA, 29. Hyakutake H, Sakai H, Yogi Y, Tsuda R, Minami Y, Yushita Y, Kanetake H, Baker A, et al.: Two variants on 17 confer prostate cancer Nakazono I, Saito Y: Beta-microseminoprotein immunoreactivity as a risk, and the one in TCF2 protects against type 2 diabetes. Nat Genet new prognostic indicator of prostatic carcinoma. Prostate 1993, 2007, 39(8):977-983. 22(4):347-355. 8. Thomas G, Jacobs KB, Yeager M, Kraft P, Wacholder S, Orr N, Yu K, 30. Ohnuma S, Miura K, Horii A, Fujibuchi W, Kaneko N, Gotoh O, Nagasaki H, Chatterjee N, Welch R, Hutchinson A, et al.: Multiple loci identified in a Mizoi T, Tsukamoto N, Kobayashi T, et al.: Cancer-associated splicing genome-wide association study of prostate cancer. Nat Genet 2008, variants of the CDCA1 and MSMB genes expressed in cancer cell lines 40(3):310-315. and surgically resected gastric cancer tissues. Surgery 2009, 9. Eeles RA, Kote-Jarai Z, Giles GG, Olama AA, Guy M, Jugurnauth SK, 145(1):57-68. Mulholland S, Leongamornlert DA, Edwards SM, Morrison J, et al.: Multiple Harries et al. BMC Cancer 2010, 10:315 Page 12 of 12 http://www.biomedcentral.com/1471-2407/10/315

31. Harries L, Brown J, Gloyn AL: Species-specific differences in the expression of the HNF1A, HNF1B and HNF4A genes. PLoS One 2009, 4(11):e7855. 32. Terasawa K, Toyota M, Sagae S, Ogi K, Suzuki H, Sonoda T, Akino K, Maruyama R, Nishiwara N, Imai K, et al.: Epigenetic inactivation of TCF2 in ovarian cancer and various cancer cell lines. Br J cancer 2006, 94(6):914-921. 33. Rebouissou S, Vasiliu V, Thomas C, Bellane-Chantelot C, Bui H, Chretien Y, Timsit J, Rosty C, Laurent-Puig P, Chauveau D, et al.: Germline hepatocyte nuclear factor 1alpha and 1beta mutations in renal cell carcinomas. Human Molecular Genetics 2005, 14(5):603-614. 34. Sun J, Zheng SL, Wiklund F, Isaacs SD, Purcell LD, Gao Z, Hsu FC, Kim ST, Liu W, Zhu Y, et al.: Evidence for two independent prostate cancer risk- associated loci in the HNF1B gene at 17q12. Nat Genet 2008, 40(10):1153-1155. 35. Bach I, Yaniv M: More potent transcriptional activators or a transdominant inhibitor of the HNF1 homeoprotein family are generated by alternative RNA preocessing. The EMBO Journal 1993, 12(11):4229-4242. 36. Romero L, Ng L, Kirby G: Chemical Inducers of Rodent Glutathione S- Transferases Down-Regulate Human GSTA1 Transcription through a Mechanism Involving Variant Hepatic Nuclear Factor 1-C. Mol Pharmacol 2006, 70(1):277-286. 37. Eiberg H, Troelsen J, Nielsen M, Mikkelsen A, Mengel-From J, Kjaer KW, Hansen L: Blue eye color in humans may be caused by a perfectly associated founder mutation in a regulatory element located within the HERC2 gene inhibiting OCA2 expression. Hum Genet 2008, 123(2):177-187.

Pre-publication history The pre-publication history for this paper can be accessed here: http://www.biomedcentral.com/1471-2407/10/315/prepub

doi: 10.1186/1471-2407-10-315 Cite this article as: Harries et al., Alterations in LMTK2, MSMB and HNF1B gene expression are associated with the development of prostate cancer BMC Cancer 2010, 10:315